The ability to differentiate brain abscess, brain tumor (primary and metastatic), and cerebral vascular accident (CVA) using the patient's age, ten ring lesion image-parameters, and two mathematical algorithms (Bayes theorem and linear discriminant function analysis) was evaluated. From the eleven patients/image parameters, a subset of six parameters was found which could partially separate the patients having those three diseases in which ring lesions occurred. The correctness of the individual disease classifications was as follows: abscess (84% accuracy), tumor (96% accuracy), and CVA (17% accuracy). The ordered list of parameters which were found to be best for separating the diseases were 1) ring thickness variability, 2) patient's age, 3) outside ring diameter, 4) average value of CT numbers in ring center, 5) maximum ring thickness, and 6) lesion (edema)-to-ring ratio. The overall classification accuracy was 86%. Three additional subsets of six parameters were identified which were disease-specific; each subset, however, was slightly different for each disease. These data are useful both for diagnosis and for identifying lesion parameters which are disease-specific and therefore deserving of further experimental or observational analysis.